Keysight 34470A bug + Severe problems with BenchVue DMM

[PIILIx-Design]

Today it boiled over with some severe problems with Keysight products...... just want to warn you about some possible issues.

Keysight 34470A has one bug I have tried Keysight to correct it without any success. The local Finnish and Swedish support seems to be totally useless and they can't make anything about this (tried 6 months). This DMM has severe current  measurement waveform distortion issue if you overrange part of the signal. It has 1:10 current ranges and sometimes you have to saturate some part of signal to see the lower parts of the signal. This however causes the hardware or software integrator etc to saturate so severely that the signal stays over the real value long time after signal has come again to measurable range (see picture). This may cause some misleading conclusions when testing.

BenchVue is quite a sad story in my current Windows 10 PC. I was so stupid and didn't test it before purchasing it and now I have a pile of non-functional crap on my PC. First version of BenchVue DMM worked quite ok but it has amazing amount of just ridiculous bugs in it. For example in waveform it showed to measurement at the same time stamp (ie. infinite rise time). After a update this was removed but much more bugs was introduced. This time software scaled time scale quite randomly 1:2. This caused me one day delay in project because I tried to find bug in software...... In every measurement session this software created some tens of errors because software is constantly sending wrong and illegal commands to the instrument. After one automatic Windows 10 update this full BenchVue and DMM stopped working at all. It halts totally at start-up. Don't have time to fight anymore. Today I tried to remove the full software from my computer and now the License manager refuses to remove or transfer the license at all.... Now I have given up, I won't use any more time to try get this junk work. It is time to accept the loss of money.....

Next frustrating issue was when I purchased new/refurbished MSOX-3104T. After spending tons of money I realized that all forums are full of discussion about the flash-problems in these scopes. Now, every time I start this I have to be afraid if it doesn't start this time..... Luckily I have some old reliable Hewlett-Packards and Agilents on my lab.

As a final conclusion: It is quite sad how difficult it is to get some really good, easy to use quality stuff these days. With the nearly unlimited technical capabilities companies are making low quality crap.

[TheSteve]

I'm not aware of a single issue with the 3000t series of scopes with respect to flash (nand) memory corruption. It was an issue with the 2000/3000a series a while back but was resolved with a firmware update. So don't worry about it.
As to your 34470a issues have you contacted Keysight in the US or posted in the Keysight support forum?

[2N3055]

Today it boiled over with some severe problems with Keysight products...... just want to warn you about some possible issues.

Keysight 34470A has one bug I have tried Keysight to correct it without any success. The local Finnish and Swedish support seems to be totally useless and they can't make anything about this (tried 6 months). This DMM has severe current  measurement waveform distortion issue if you overrange part of the signal. It has 1:10 current ranges and sometimes you have to saturate some part of signal to see the lower parts of the signal. This however causes the hardware or software integrator etc to saturate so severely that the signal stays over the real value long time after signal has come again to measurable range (see picture). This may cause some misleading conclusions when testing.

BenchVue is quite a sad story in my current Windows 10 PC. I was so stupid and didn't test it before purchasing it and now I have a pile of non-functional crap on my PC. First version of BenchVue DMM worked quite ok but it has amazing amount of just ridiculous bugs in it. For example in waveform it showed to measurement at the same time stamp (ie. infinite rise time). After a update this was removed but much more bugs was introduced. This time software scaled time scale quite randomly 1:2. This caused me one day delay in project because I tried to find bug in software...... In every measurement session this software created some tens of errors because software is constantly sending wrong and illegal commands to the instrument. After one automatic Windows 10 update this full BenchVue and DMM stopped working at all. It halts totally at start-up. Don't have time to fight anymore. Today I tried to remove the full software from my computer and now the License manager refuses to remove or transfer the license at all.... Now I have given up, I won't use any more time to try get this junk work. It is time to accept the loss of money.....

Next frustrating issue was when I purchased new/refurbished MSOX-3104T. After spending tons of money I realized that all forums are full of discussion about the flash-problems in these scopes. Now, every time I start this I have to be afraid if it doesn't start this time..... Luckily I have some old reliable Hewlett-Packards and Agilents on my lab.

As a final conclusion: It is quite sad how difficult it is to get some really good, easy to use quality stuff these days. With the nearly unlimited technical capabilities companies are making low quality crap.

You are correct, BenchVue is of horrible quality and should be free.. It used to be.. The way it is now, they would have to pay ME to use it..

I apologize but I have to adamantly disagree with some of your other statements. You are using an instrument deliberately outside it's specifications, overloading it and possibly even damaging it and than state Keysight is crap because, while you are abusing an instrument, measurements are not in specifications...

It is almost the same as saying: "Keysight is making crap, just the other day I burned my display trying to put out a cigarette, for the money I would expect them to put glass there, so I can use it as an ashtray....". Once you are using it outside specs and intended purpose, all bets are off.

Flash rot is a documented problem with 2000/3000A and 4000A series... Since 3000T is basically based on 4000A series, in theory it might have problems too.. But as Steve said, we have yet to see one that was bricked.
I don't know if that is because they fixed underlying hardware source of the problem, or because of those scopes having newer firmware that has a fix in it in a first place.

Regards,

[maxwell3e10]

I've never tried to check DMM digitizers for overload recovery, but just the other day I run into issues with overload recovery of oscilloscopes.

Sometimes it is important to see the tail end of a large signal in detail. Tektronix, Keysight and Micsig scopes gave me totally different signal shapes. Still have to figure out which one is right.

[HighVoltage]

This DMM has severe current  measurement waveform distortion issue if you overrange part of the signal. It has 1:10 current ranges and sometimes you have to saturate some part of signal to see the lower parts of the signal. This however causes the hardware or software integrator etc to saturate so severely that the signal stays over the real value long time after signal has come again to measurable range (see picture). This may cause some misleading conclusions when testing.

I have used the 34470A for detailed current measurements and have not come across any problems at all in any current range. To the contrary, it measures accurately in to the nA range.
 
Why are you over-ranging this instrument?
Can you share some pictures of your problems?

I agree with you on BenchVue, even some of my clients / customers have stopped using BenchVue and saw it as a waste of money because of far too many problems.

[bson]

Why are you over-ranging this instrument?

Because he's trying to measure a small quiescent current in the presence of a large load spike?

[2N3055]

Why are you over-ranging this instrument?

Because he's trying to measure a small quiescent current in the presence of a large load spike?

On a 10A range you can resolve uA currents.... It's a7.5 digit instrument.
Even so, you can clamp overcurrent  event since you are not measuring it anyways..
Or just have a capacitor that will integrate pulse..

[PIILIx-Design]

This overload recovery issue was found by accident when I was studying the use of this 34470A. I was measuring very difficult current waveform and get very strange results before I realized that 34470A has quite high value shunt resistor on lower current ranges (1kohm 1uA-100uA, 100R 1mA, 2R7 10-100mA). Because of this resistance it causes significant distortion to the current waveform in high dynamic signal. See pictures of the current waveforms, 10mA range looks real but when measured at 1mA range signal is totally different. Device I tested is lithium coil fed humidity meter, switch-mode power supply 3->5V, measures 1 time / min. Peak current during measurement ~2.5mA, during sleep ~2uA, average ~3.5uA (1:1250 dynamic). To get accurate reading from this kind of current signal was rather complicated. I had to measure every part of the signal with different scale and combine the results in excel.

Some comments.......
---------------
On a 10A range you can resolve uA currents.... It's a7.5 digit instrument.
Even so, you can clamp overcurrent  event since you are not measuring it anyways..
Or just have a capacitor that will integrate pulse..
---------------
This is the most common error people make. Resolution is totally different than accuracy. 34470A has plenty of resolution but absolute accuracy in higher current ranges is in fact quite low. From the specification you can see that 10A range accuracy is specified (1 year spec) 0.120% reading + 0.010% of range. This gives error of +-1mA when measuring very low current. So it is totally useless to measure 1uA with 10A range. Even with the 10mA range (minimum not clipped) error may be 2uA. This is far too much related to the average current of 3.5uA of the device (+-57% error)

In this case capacitor can't be used because I need RMS values. Lithium coil cell life calculation is strongly dependent on current waveform, not only average.

About the flash issues in scopes. If they have used the same flash-IC' in the 3104T scope there is timebomb even if there is some firmware enhancements. When too many flash cell are corrupted the device will in some day stop working. It is a crude fact. Hopefully they have used different flash..... Test instruments are normally used as long they work, because there is normally no reason to update them. 20-30 years is quite normal life for scopes and power supplies etc. They are long time investments, especially in one-man companies like mine.

Now I need to continue with my projects. Just wanted you to know you might meet some overload recovery issues with your 34470A

[AndyC_772]

What's the time scale over which this problem occurs? How long does it take after going over-range before the instrument reads accurately again?

[PIILIx-Design]

In those latest excel pictures there are no any problem in the 34470A. The waveform corruption is because the full system is behaving differently after I have inserted the internal shunt resistor (100R?) of the 34470A to the system. Because of switch-mode power supply (=negative resistance) this serial resistance between lithium coil battery and the humidity meter causes input voltage to drop. This causes input current to rise and it distort the measurement. This was just example what can go wrong when measuring high dynamic signals. Ideal current measurement adds zero resistance to the system.

The overload recovery is pictured in my first message picture. It happens when you use low frequency (~10Hz) and use ~2-3 x the maximum signal. For me it looks like they have software filter and they don't stop the integrator part of the function when the signal is clipped. This causes integrator to grow and after signal is again in displayed region it takes long time before software filter output comes to sensible level. This is only a wild guess and there maybe some issues in the hardware also.  This was however first time in my career (30 years) when i met this kind of behavior in any test instrument. All older Agilent scopes etc seems to take all kind of punishment without any problem.

[maxwell3e10]

Measuring currents without affecting them is always a little tricky. For 10 mA range you can use a op-amp current to voltage converter.  That way the load remains at virtual ground. Dave (the owner of this site)  sells a little box muCurrent that does something similar (using a very small shunt resistor and a zero-drift x1000 amplifier) this site (although it seems to be out of stock). But you can put it together yourself with one op-amp and one resistor.

34470 has a separate digitizer mode, where a simple ADC is used for measurements instead of a fancy integrating one. It may have a better overload recovery response. Also, autoranging needs to be turned off, of course.

As far as DMMs go, there hasn't really been any real hardware progress in the last 30 years, so an old DMM like HP3457 can do almost as much as newer ones except for a fancy display.

[Testtech]

Could you use a current probe and a scope?